The below are my thoughts. People are free to develop their own.

In my view, we probably need both on and off chain solutions as this would widen BTC usability cases. I fail to see how off-chain solutions would work on top of an unreliable and restricted main chain. Even the lightning network white paper raises the need for on-chain growth.

So I personally would vote 'yes' to a blocksize increase as this is the simplest solution. Other people may consider solutions such as extension blocks to be a better way forward.

As for segwit, I need to be convinced that segregated witness blocks are the simplest solution. I suspect it's benefits could be achieved simply by new transaction format types, (similar to flextrans). So at the moment I am inclined to vote 'no', but I am open to be convinced with greater technical arguments. (And implementing a hard fork as a soft fork with software engineering hacks due to Fear Of Hark Fork is not something that would sway me.)